NCSA NEWS

During World War II, researchers at Edgewood Arsenal in Maryland found something they didn't expect to find. They were curious about gelled gasoline, the thick, jelly-like substance used in flame throwers--formed by adding a chemical thickener, such as aluminum soap, to gasoline. The researchers noticed that at high flow rates the thickened gas flowed with less pressure, as much as 50 percent lower drag, than the liquid.

At the time, due to wartime security, this intriguing finding went unreported. In 1949, a British chemist, B. A. Toms, announced his similar, independent finding that dissolving small quantities of heavy, long-chain molecules--polymers--in solution reduced drag during turbulent flow through a tube by up to 70 percent. Since the 1960s, a large body of experimental knowledge has built up on this phenomenon--sometimes called "the Toms effect." It has been applied in the oil industry: in fluids pumped at high pressure to fracture oil-bearing rock formations, and in the Alaskan pipeline, where mixing polymers with crude oil cuts pumping cost in half.

What, exactly, is going on here? Centuries of engineering experience, not to mention common sense, tell us that thickened "heavy" fluids should flow with more resistance than "thin" liquids. The experimental work has fostered a good deal of theorizing about this quirky phenomenon, but in many respects it remains an enigma. What characteristics of the polymer additives govern the effect? How can it be controlled and used in other applications? Despite almost half a century of research, says Antony Beris, "details are sketchy and direct evidence for any specific mechanism is still lacking"--which is where supercomputing enters the picture.

A professor of chemical engineering at the University of Delaware, Beris has taken on the challenge of building an accurate computational model of this phenomenon. He appears to have solved some vexing problems (numerical instabilities) that defeated prior attempts at modeling these flows. His recent work--in collaboration with former graduate student R. Sureshkumar (now on the faculty at Washington University in St. Louis)--is the first successful simulation of polymer-induced drag reduction "from first principles." Starting from scratch, without experimental data--only the governing mathematical expressions and a theoretical model of polymers, he obtained results that match observed data about these thickened fluids. More importantly, his results contribute at the theoretical level--by offering new support for an explanation of the reduced drag that relates the molecular characteristics of polymers and the macroscopic behavior of turbulent flow.